obra/superpowers

What it solves

It addresses the tendency of AI coding agents to jump immediately into writing code without a proper plan, often leading to poor design, lack of tests, and deviations from the user's actual goals.

How it works

Superpowers provides a set of composable skills and instructions that act as a software development methodology for agents. Instead of immediate coding, it forces a structured workflow: brainstorming and spec refinement, creating a detailed implementation plan, and then executing that plan using a subagent-driven-development process. It integrates with various AI coding harnesses (like Claude Code, Cursor, and GitHub Copilot CLI) as a plugin or extension to automate these behaviors.

Who it’s for

Developers using AI coding agents who want more reliable, professional-grade software engineering outcomes through structured processes like TDD and systematic debugging.

Highlights

• Subagent-Driven Development: Uses specialized subagents to execute tasks and perform two-stage reviews for spec compliance and code quality.
• Strict TDD Enforcement: Mandates a RED-GREEN-REFACTOR cycle where tests are written before any production code.
• Systematic Debugging: Implements a four-phase root cause process rather than ad-hoc guessing.
• Broad Compatibility: Supports a wide range of tools including Claude Code, Cursor, Gemini CLI, and GitHub Copilot CLI.